The Theories and Determinants of Teacher Attrition and Retention
By Tuan D. Nguyen
Dissertation
Submitted to the Faculty of the Graduate School of Vanderbilt University
in partial fulfillment of the requirements for the degree of
DOCTOR OF PHILOSOPHY in
Leadership and Policy Studies
June 30, 2018 Nashville, Tennessee
Approved:
Matthew G. Springer, Ph.D.
Christopher A. Candelaria, Ph.D.
Joseph F. Murphy, Ph.D.
Emily E. Tanner-Smith, Ph.D.
Copyright © 2018 by Tuan D. Nguyen All Rights Reserved
ACKNOWLEDGEMENTS
I am indebted for the support and mentoring from the faculty in Peabody College’s Department of Leadership, Policy and Organizations. In particular, Brent Evans has been an extraordinary advisor, mentor, collaborator, and friend. Our weekly meetings have often been the most productive and enjoyable time I have had in the past few years. As committee chairs, Matt Springer and Chris Candelaria provided routine and supportive feedback that greatly sharpened my own thinking and writing.
I am grateful for the support from Joe Murphy, Marisa Cannata, Will Doyle, and Emily Tanner-Smith. They have taught me how to think like a researcher and the importance and potential value of research to improve educational outcomes for both teachers and students. I am also thankful for the support of my cohort, C.J., Jenna, Laura, Sarah, Susan, and Sy.
I am thankful for my parents and their sacrifices. I am here today because of the support and guidance they have provided me over the years.
Finally, I would like to acknowledge the support of my family, Kate, Mira, and Ella, who put everything in perspective. It has been wonderful and challenging to learn and develop a whole new set of skills over the past five years, and it is all the sweeter that I have been able to do it side-by-side with Kate as she earns her own doctoral degree. Kate, thank you for being there every day and for your patience and kindness.
TABLE OF CONTENTS
Page
ACKNOWLEDGEMENTS ... iii
LIST OF TABLES ... vi
LIST OF FIGURES ... ix
Chapter I. Introduction ... 1
II. A Conceptual Framework Of Teacher Attrition And Retention: A Systematic Review Of The Empirical Literature And Insights From The Employee Turnover Literature ... 5
Introduction ... 5
Examining Prior Conceptual Frameworks ... 8
Data and Methods ... 19
Results ... 22
Discussion and Conclusion ... 37
III. The Determinants Of Teacher Attrition And Retention: A Meta-Analysis Of The Literature ... 41
Introduction ... 41
Motivating the Current Study ... 45
Conceptual Framework ... 47
Data & Methods ... 58
Analytic Strategy ... 64
Results… ... 69
Discussion and Conclusion ... 90
IV. The Factors Of Teacher Attrition And Retention: Evidence From Repeated Cross- Sectional National Data ... 96
Introduction ... 96
Literature Review... 100
Data and Methods ... 105
Results ... 111
Discussion and Conclusion ... 128
REFERENCES ... 134
Appendix
A. Chapter II Appendix... 162 B. Chapter III Appendix ... 164 C. Chapter IV Appendix ... 176
LIST OF TABLES
Table Page
Chapter II. A Conceptual Framework Of Teacher Attrition And Retention: A Systematic Review Of The Empirical Literature And Insights From The Employee Turnover Literature
Table 1: Categories and determinants of teacher attrition and retention ... 14
Table 2: Results by database ... 21
Appendix Table 1: Coding and descriptions of determinants ... 162
Chapter III. The Determinants Of Teacher Attrition And Retention: A Meta-Analysis Of The Literature Table 1: Categories and determinants of teacher attrition and retention ... 50
Table 2: Results by database ... 62
Table 3. Descriptive information on the primary studies by study characteristics ... 68
Table 4: Personal correlates as determinants of teacher attrition and retention using dependent effect sizes ... 71
Table 5: School correlates as determinants of teacher attrition and retention using dependent effect sizes ... 76
Table 6: External correlates as determinants of teacher attrition and retention using dependent effect sizes ... 81
Table 7. Bivariate meta-regression results ... 90
Appendix Table 1: Coding and descriptions of determinants ... 164
Appendix Table 2: Study quality and risk of bias considerations ... 166
Appendix Table 3: Determinants of teacher attrition and retention using independent effect sizes ... 167
Appendix Table 4: Determinants of teacher attrition and retention with robust variance estimation ... 169
Appendix Table 5: Determinants of teacher attrition and retention without hazard ratios estimates ... 170 Appendix Table 6: Determinants of teacher attrition and retention using “Leavers only”
estimations (without leavers and switchers combined as one group) ... 172 Appendix Table 7: Determinants of teacher attrition and retention for teachers leaving their current school (Leavers, Leavers and Switchers, and Switchers)………171
Chapter IV. The Factors Of Teacher Attrition And Retention: Evidence From Repeated Cross- Sectional National Data
Table 1. Descriptive Statistics of teacher characteristics, teacher qualifications, school organizational characteristics, school resources, and student body characteristics for public school teachers ... 109 Table 2. National rate of attrition for public school teachers ... 112 Table 3. Multinomial logistic regression estimates of the influence of teacher characteristics on movers and leavers for public school teachers ... 112 Table 4. Multinomial logistic regression estimates of the influence of teacher qualifications on movers and leavers for public school teachers ... 114 Table 5. Multinomial logistic regression estimates of the influence of school organizational characteristics on movers and leavers for public school teachers ... 117 Table 6. Multinomial logistic regression estimates of the influence of school resources on movers and leavers for public school teachers ... 118 Table 7. Multinomial logistic regression estimates of the influence of student body
characteristics on movers and leavers for public school teachers ... 119 Table 8. Saturated models of multinomial logistic regression estimates of the determinants of movers and leavers for public school teachers ... 120 Table 9. Changes over time for select variables (saturated pooled models) ... 127 Appendix Table 1. Definitions of School Organizational Characteristics Used in Regression Analysis... 176 Appendix Table 2. Descriptive Statistics of teacher characteristics, teacher qualifications, school organizational characteristics, school resources, and student body characteristics for public school teachers including charter schools ... 177
Appendix Table 3. Marginal effect estimates from multinomial logistic regression models of the influence of teacher characteristics on movers and leavers for public school teachers ... 178 Appendix Table 4. Marginal effect estimates from multinomial logistic regression models of the influence of teacher qualifications on movers and leavers for public school teachers ... 179 Appendix Table 5. Marginal effect estimates from multinomial logistic regression models of the influence of school organizational characteristics on movers and leavers for public school
teachers ... 180 Appendix Table 6. Marginal effect estimates from multinomial logistic regression models of the influence of school resources on movers and leavers for public school teachers ... 181 Appendix Table 7. Marginal effect estimates from multinomial logistic regression models of the influence of student body characteristics on movers and leavers for public school teachers ... 182 Appendix Table 8. Marginal effects from saturated models of multinomial logistic regression estimates of the determinants of movers and leavers for public school teachers ... 183 Appendix Table 9. Saturated models of multinomial logistic regression estimates of the
determinants of movers and leavers for public school teachers including charter schools ... 185 Appendix Table 10. Saturated models of multinomial logistic regression estimates of the
determinants of movers and leavers for public school teachers without sampling weights ... 189
LIST OF FIGURES
Figure Page
Chapter II. A Conceptual Framework Of Teacher Attrition And Retention: A Systematic Review Of The Empirical Literature And Insights From The Employee Turnover Literature
Figure 1. Conceptual Framework of Teacher Attrition and Retention ... 13 Chapter III. The Determinants Of Teacher Attrition And Retention: A Meta-Analysis Of The Literature
Figure 1. Conceptual Framework of Teacher Attrition and Retention ... 51 Figure 2. Flow diagram ... 61 Figure 3. Forest plot for effect estimates of gender (male) on teacher attrition from primary studies. ... 73 Figure 4. Contoured enhanced funnel plot of gender (male) and attrition ... 88
CHAPTER I INTRODUCTION
Teachers represent a critical part of public education and there is compelling interest in retaining teachers, particularly for schools in dire needs. This has led researchers and policy makers to develop strategies to recruit and retain effective teachers, highlighting the importance of knowing the determinants of teacher attrition and retention. As such, this dissertation seeks to contribute to the scholarly understanding of the determinants of teacher attrition and retention in a series of three studies, demarcated as chapters II, III, and IV. The studies are strongly connected and each one informs the work of the other.
The first study, chapter II of the dissertation, develops a comprehensive conceptual framework of teacher attrition and retention. The two most cited and seminal works in the area of teacher attrition and retention are the Guarino, Santibanzes, and Daley (2006) and Borman and Dowling (2008) studies. While both studies present their own conceptual framework, the frameworks were constructed without much guidance from the employee turnover literature.
Moreover, their frameworks are based solely on the factors that they found in their narrative review and meta-analysis. Last but not least, there has been substantial development in the area of teacher attrition and retention in the last twelve years since these two seminal works were published.
This study then makes several contributions to the study of teacher attrition and retention.
Building on these two seminal works, I present a comprehensive conceptual framework of teacher attrition and retention, which is guided by broader employee turnover literature and supported by the empirical literature. Using updated systematic review methods, leveraging the broader employee turnover literature, and adding in more than ten additional years of research on
the determinants of teacher attrition and retention, particularly in light of the proliferation of “big data” and its use in providing new results and causal estimates, I categorized the determinants into nine subcategories grouped under three primary categories of personal correlates, school correlates, and external correlates. Personal correlates include teacher characteristics and teacher qualifications; school correlates include school organizational characteristics, school resources, student body characteristics, and relational demography; and external correlates include school improvement, work force, and accountability. For each category, I discuss the empirical results and their implications for attrition and retention and I also highlight the gaps in the empirical literature and the possible policy levers to positively influence the teacher work force. In short, through the systematic review and synthesis of the literature I created a framework that can be used to study and advance the field’s knowledge on teacher attrition and retention in addition to synthesizing the results of nearly forty years of research and providing guideposts for future research in this area of scholarly study.
The second study, chapter III of the dissertation, is a meta-analysis of teacher attrition and retention using the conceptual framework developed in the first study. I improve upon the Borman and Dowling (2008) study by expanding the search from 1980 to 2017, adding in more than a decade of research and development, which is particularly meaningful with the inclusion of studies involving teacher evaluation, teacher merit pay, NCLB, principal effectiveness, teacher-principal race/gender matching, teacher-student race matching, comprehensive school reform, and research-practice partnership. Moreover, the proliferation of longitudinal data systems has allowed researchers to empirically improve our understanding of the teacher labor market dynamics, which substantially advances the field’s knowledge of teacher attrition and retention in a meta-analysis. The study also makes methodological improvements using current
best meta-analytic practices, paying close attention to the grey literature search, publication bias, data analysis, presentation of results and robustness of findings. The updates and improvements proposed in this meta-analysis should enhance the scholarly understanding of what drives teacher attrition and retention and provide the most up-to-date comprehensive review of the field’s empirical knowledge on teacher attrition and retention.
The third study, chapter IV of the dissertation, is a quantitative study of the determinants of teacher attrition and retention using over-time cross-sectional national data. In the current literature the factors determining retention and attrition come from a large number of studies that vary substantially from each other with respect to time period, location, sample, sampling strategy, and methodology. Empirically, each factor has been found to have an association with teacher retention and attrition, but no study has analyzed the extent to which these relationships hold when these factors are studied simultaneously. Another limitation is that the majority of the studies have analyzed the relationship between the factors and teacher retention and attrition in a given year or a single district or state. Consequently, there is substantial room for contribution to the research on the factors of teacher retention and attrition.
This study analyzes whether and how the relationships of these categories of factors and teacher retention and attrition hold true nationally and over time using four waves of the Schools and Staffing Survey (SASS) and its supplement, the Teacher Follow-Up Survey (TFS). Using these data, I make several contributions to the study of teacher attrition and retention. First, the use of longitudinal nationally representative data instead of short-term district- or state-level data to study teacher turnover provides a more thorough picture of turnover nationally and over time.
Second, the data from SASS have detailed comprehensive information on teacher characteristics, teacher qualifications, school organizational characteristics, school resources, and student body
characteristics, which is ideal in studying how these factors influence teacher turnover as it alleviates some concerns of omitted variable bias. Third, I am able to differentiate between movers and leavers, and I find that factors that influence movers may not influence leavers and vice versa, which substantially adds to the empirical knowledge of the factors of teacher attrition and retention. Fourth, I find that, while the influences of most factors remain stable through time, there are some variables whose influences have changed over the twelve-year time frame.
In short, the first two studies provide a framework that can be used to study and advance the field’s knowledge on teacher attrition and retention in addition to synthesizing the results of nearly forty years of research and providing guideposts for future research, while the third study examines whether and how factors of teacher retention and attrition hold true nationally and over time, and how factors that influence leavers may not influence movers in the same way. Taken together, the three studies in this dissertation add substantial breadth and depth to the study of and policy discourse on teacher attrition and retention.
CHAPTER II
A CONCEPTUAL FRAMEWORK OF TEACHER ATTRITION AND RETENTION: A SYSTEMATIC REVIEW OF THE EMPIRICAL LITERATURE AND INSIGHTS FROM
THE EMPLOYEE TURNOVER LITERATURE Introduction
Teacher labor markets and teacher shortages have strong implications for learning outcomes and equity for students. Teachers are the foundation of public education, an integral part of a democratic society. As such, the general public and educators care about issues of equity and productivity in schools, and policy makers have spent a considerable amount of time working to ensure that classrooms are staffed with qualified teachers (Hanushek, Kain, & Rivkin, 2004; Ingersoll & Smith, 2003; Loeb, Kalogrides, & Beteille, 2012). For instance, No Child Left Behind (NCLB) and Every Student Succeeds Act (ESSA) are two pieces of federal legislation that emphasize the importance of teachers and the placement of qualified teachers in every classroom. Substantial evidence indicates that an important facet of the large variations in quantity and quality of the teacher workforce among schools and districts is the teacher attrition rates in certain schools and districts, and scholars have concluded that we need to learn more about teacher labor markets so that we can better address the uneven distribution of quantity and quality of teachers (Feng & Sass, 2017a; Ingersoll, 2001; Guarino, Santibanez, & Daley, 2006;
Lankford, Loeb, & Wyckoff, 2002).
Variations in teacher quantity and quality can be explained by multiple factors such as teacher preferences and district hiring practices (Engel & Cannata, 2015; Lankford, Loeb, &
Wyckoff, 2002). These factors affect how teachers are sorted differentially across states, districts and schools, and substantial research has shown that the most disadvantaged schools, schools that need effective teachers the most, are often the hardest to staff and often face high teacher
turnover (Allensworth, Ponisciak & Mazzeo, 2009; Boyd et al., 2011; Ingersoll, 2001; Guarino, Santibañez, & Daley, 2006). In addition, high teacher turnover is also costly to schools and districts (Barnes, Crowe, & Schaefer, 2007). In short, teachers represent a critical part of public education and there is compelling interest in retaining quality teachers, particularly for disadvantaged schools. This has led researchers and policy makers to develop strategies to recruit and retain effective teachers, highlighting the importance of the determinants of teacher attrition and retention.
What can explain why teachers stay or go? There are few comprehensive conceptual frameworks based in both the employee turnover literature and empirical research in education that explain the determinants of teacher attrition and retention. Individual studies of teacher attrition and retention often present a tailored framework that focuses on the specific factors that are examined in that study. In terms of critical and systematic reviews of teacher attrition and retention, the two most cited and seminal works are the Guarino, Santibañez, and Daley (2006) and Borman and Dowling (2008) studies. While both studies made substantial contributions to the study of teacher attrition and retention with their reviews and conceptual frameworks, there are limitations to each study that can be addressed. In particular, while Guarino, Santibañez, and Daley (2006) presented a conceptual framework based on the economic labor market theory of supply and demand, this conceptual framework was fairly broad as it encompassed both the supply and demand sides. As such, their conceptual framework does not provide a detailed framework for studying teacher attrition and retention. Relatedly, the authors only dealt with several factors that influence attrition and retention.
On the other hand, based on their systematic search of the empirical literature on teacher attrition Borman and Dowling (2008) presented five categories of attrition and retention factors
with several factors under each category. However, these categories were grouped together without guidance from the prior literature, particularly the broader employee turnover literature.
Moreover, both of these frameworks are based solely on the factors identified through their reviews without consideration of theoretical and not-yet-available factors that could drive attrition and retention, which is one area this study seeks to advance. Furthermore, a substantial amount of work on teacher attrition and retention has been conducted since these two seminal studies were published, particularly in the areas of evaluation, value-added measures, and merit pay. Relatedly, there have been significant methodological improvements in the field of systematic reviews that allow for more thorough and robust searches to find eligible studies.
Moreover, the development and proliferation of large longitudinal data systems, “big data,” in the last ten years have allowed researchers to empirically advance our understanding of teacher attrition and retention, especially in terms of providing causal estimates. As such, these developments provide new insights and opportunities to create and develop a more comprehensive conceptual framework of teacher attrition and retention.
In sum, there is strong need for a comprehensive conceptual framework of teacher attrition and retention that is guided by employee turnover literature and supported by empirical evidence attending to recent development and insights. Thus, to advance the literature on teacher attrition and retention I develop a conceptual framework of teacher attrition and retention based on a systematic search of the empirical literature on teacher attrition and retention and guided by the existing literature on employee turnover of which teacher turnover is a subset. In particular, the research questions I address are:
(1) What are the conceptual and empirical determinants of teacher attrition and retention?
(2) Are there gaps in the empirical research on teacher attrition and retention? If so, what are they?
(3) What are some policy levers that can be used to affect teacher attrition and retention?
In reviewing and synthesizing the literature on teacher attrition and retention, I follow the recommended processes of Hallinger (2014), Moher et al. (2009), and Murphy (2008). As such, the rest of the paper is structured as follows. First, I discuss previous conceptual frameworks of teacher attrition and retention as well as the employee turnover literature that guides my construction of a comprehensive conceptual framework of teacher attrition and retention. Then I describe the procedures used to conduct my systematic review of this extensive body of empirical and theoretical scholarship. Based on this systematic review, I discuss the findings of my search in light of my conceptual framework, which contains three primary categories of personal correlates, school correlates, and external correlates. In the discussion section I highlight the new areas of development and empirical gaps in the literature on teacher attrition and retention as well as promising policy levers that may be able to positively impact the composition of the teacher work force.
Examining Prior Conceptual Frameworks
Guarino, Santibañez, and Daley (2006), one of the most cited works in the area of teacher attrition and retention, provided a conceptual framework based on the economic labor market theory of supply and demand. With this broad frame in mind, they examined the literature related to teacher entry, mobility and attrition and summarize the prominent themes in these areas. For instance, in terms of the characteristics of individuals who enter teaching, they found that gender, race/ethnicity, ability, and psychological and family-related factors all contribute to the
composition of who enters the profession. Relatedly, they found that age, experience, gender, race/ethnicity, ability, field or specialization, qualifications, and psychological factors influence teacher decisions to leave teaching. In particular, they found teacher attrition is high for teachers in their first few years of teaching, that minority teachers tend to have lower attrition rates than White teachers, teachers in math and science are more likely to leave than teachers in other fields, and teachers with higher measured academic ability and female teachers are also more likely to leave. Furthermore, they discussed the characteristics of districts and schools that successfully recruit and retain teachers, and they concluded that schools with higher proportions of minority, low-income and low-performing students tend to have higher attrition rates. Lastly, they categorized and examined policy levers (compensation policies, pre-service policies, and in- service policies) that may be able to positively affect the composition of the teacher work force.
For instance, they found that teachers respond positively to higher salaries and that attrition rates are lower in schools that provide mentoring and induction programs. In short, using the broad framing of economic labor market theory of supply and demand Guarino, Santibañez, and Daley (2006) summarized some of the driving forces of teacher attrition and retention.
Building on Guarino, Santibañez, and Daley (2006), Borman and Dowling (2008) conducted a meta-analysis and narrative review of the research on teacher attrition and retention.
Across 34 studies, they found over 60 factors that are empirically associated with teacher attrition and retention. They organized these factors into five categories: teacher characteristics, teacher qualifications, school organizational characteristics, school resources, and student body characteristics. Teacher demographic variables facilitate our understanding of how teacher background characteristics such as age, gender, race/ethnicity, and marital status, influence attrition and retention. Teacher qualifications include teacher training, certification, teaching
experience, teacher ability, and field or specialty area. School organizational characteristics are the characteristics of the school such as the urbanicity, size, secondary versus elementary level, and work environment. School resources drill down further to include characteristics such as average class size and teaching materials. Lastly, student body characteristics describe the composition of the student body, which includes broad areas such as the school’s socioeconomic composition, student achievement level, and the racial/ethnical composition of the school. For each of these five categories, Borman and Dowling (2008) presented their meta-analytic results.
This study included a more thorough examination of the factors that may influence attrition and retention than the work by Guarino, Santibañez, and Daley (2006). However, Borman and Dowling (2008) constructed these five categories based on how they seem to group together and without guidance from the broader employee turnover literature. The five categories certainly are reasonable categories but they do not provide a conceptual framework that situates their findings and that could be used to point to gaps in the literature. Consequently, I build upon these two seminal studies by conducting a systematic search of the empirical literature on teacher attrition and retention, and by developing a classification scheme for identified determinants of teacher attrition and retention that is guided by existing literature on employee turnover at large. More specifically, I focus on the factors that influence whether teachers exit the system, i.e., leave teaching or leave the state or the data set where they taught, not on whether teachers switch schools.1 In addition to updating the field’s knowledge of the determinants of teacher attrition and retention with more than ten years of additional research since Borman and Dowling (2008)2 and Guarino, Santibañez, and Daley (2006), I provide a conceptual framework that organizes
1 For ease of reading, I also use leave teaching when referring to teachers exiting the system as many studies do.
prior findings with insights from the broader employee turnover literature, which I describe below.
A classic study that has influenced the scholarly study of employee turnover is a meta- analysis and review by Cotton and Tuttle (1986). In this study, Cotton and Tuttle (1986) analyzed the determinants of employee turnover and organize them into three large categories, or correlates, of turnover: (1) personal correlates, (2) work-related correlates, and (3) external correlates. Personal correlates are variables and characteristics of the employees and their relationship with turnover such as age, gender, education, marital status, number of dependents, and ability. Work-related correlates include variables that are associated with the work place such as job satisfaction, salary satisfaction, and organizational commitment. External correlates include factors that are outside the individual or the work place such as the unemployment rate and union presence.
Similar to Cotton and Tuttle (1986), Griffeth, Hom, & Gaertner (2000) conducted a meta- analysis of the determinants of employee turnover. In their work, they organize the determinants into six categories: (1) personal characteristics, (2) satisfaction with overall job and job facets, (3) other dimensions of work experience, (4) external environment factors, (5) behavioral predictors, and (6) cognitions and behaviors about the withdrawal process. The authors also found that, like Cotton and Tuttle (1986), these factors do indeed predict turnover, and with more studies and variables available, they organized them into more detailed categories.
The most recent systematic review or meta-analysis study on the employee turnover literature was conducted by Rubenstein, Eberly, Lee, and Mitchell (2017). Building on the prior literature, Rubenstein and colleagues conducted a meta-analysis of the determinants of employee turnover. Following prior work, they categorized their findings into nine categories: (1)
individual attributes, (2) aspects of the job, (3) job attitudes, (4) newer personal conditions, (5) organizational context, (6) person-context, (7) external job market, (8) attitudinal withdrawal, and (9) employee behaviors.
Looking across these three studies and others (e.g., Maertz, Griffeth, Campbell, & Allen, 2007; Ongori, 2007; Porter & Steers, 1973), there seem to be three large categories and smaller subcategories that influence general employee turnover. First, the individual characteristics of employees such as age, gender, race/ethnicity, and qualifications are highly associated with turnover. Second, the interactions between the individual employees and the work place also matter. In this construct, we observe that variables like the physical conditions of the work place and support and collaboration among employees play a significant role in whether people leave their job. Third, there are external events or factors outside the individual and the work place, such as the external job market, that influence whether people stay or leave their job.
Adapting the findings from these prior works in employee turnover literature to teacher attrition and retention literature, I argue there are three primary categories that influence teacher attrition and retention: (1) personal correlates, (2) school correlates, and (3) external correlates.
Under personal correlates, we have teacher characteristics and teacher qualifications categories.
School correlates contains factors that describe the schools and conditions in which the teachers work, including school organizational characteristics, school resources, student body
characteristics, and relational demography categories. Lastly, under external correlates, are factors that come from federal, state, or district policies and economic factors surrounding employment such as accountability, school improvement, and work force categories. Figure 1 is the visual representation of my conceptual framework of teacher attrition and retention and Table 1 lists the specific factors under each category. Five of the nine categories are based on Borman
and Dowling (2008), while four categories, relational demography, accountability, school improvement, and work force, are proposed categories that have clear theoretical reasons or empirical evidence that may influence teacher attrition and retention, which I briefly describe below.
Figure 1. Conceptual Framework of Teacher Attrition and Retention School Correlates
School Organizational Characteristics School Resources
Student Body Characteristics Relational Demography
External Correlates Accountability School Improvement
Work Force Personal Correlates
Teacher Characteristics Teacher Qualifications
Teacher Attrition and Retention
Table 1: Categories and determinants of teacher attrition and retention
Personal correlates School correlates External correlates
Teacher characteristics
Teacher qualifications
School org characteristics
School resources
Student body characteristics
Relational demography
Accountability School improvement
Work force Age Ability (test
scores)
School size Expenditure Student achievement
Tch-princ race/gender match
Assessment impact
Mandated school reform
Employment rate
Gender Education selectivity
Urbanicity Class size Percent minority
Tch-tch race match
Teacher effectiveness
Research- practice partnership
Accession rate Race/ethnicity Graduate
Degree
Sec. vs elem.
level
Classroom assistants
Poverty Tch-student race match
Merit pay Late hiring
Marital status Certification Private, public, charter
Teaching materials
Percent IEP/LEP
Federal policies (NCLB/ESSA)
Salary Children Highly
qualified (NCLB/ESSA)
Work environment
Principal effectiveness
Retention bonus Satisfaction Internship Administrative
support
Non-teacher salary Full time
Teaching
Specialty area (STEM, SPED)
Teacher collaborations
Union
Distance to school
Experience Teacher leadership
Tenure Prior non-
teaching career experience
Professional development Induction mentoring Classroom autonomy Stay ratio
Note. The first five categories are adapted and expanded based on Borman & Dowling (2008). Stay ratio is the teacher retention rate at the school. Internship includes field placement. Teacher leadership includes teacher influence at the school level. Asssessment impact includes evaluation used for school-level decision-making. In comparison, teacher effectiveness score is measured by a composite evaluation score or value-added score.
Relational demography. A new area of development in the literature on teacher attrition and retention comes from the relational demography, and relatedly representative bureaucracy, literature (Fairchild et al., 2012; Grissom, Kern, & Rodriguez, 2015; Sohn, 2009). The thrust of the theory of relational demography is that people are influenced by the composition of other people around them. In other words, the degree of similarity between a person and others with whom they have regular contact would influence their attitudes and behaviors (Tsui & O’Reilly, 1989). The literature in this area suggests that there are positive outcomes for teachers and students when there is race or gender congruence between the teachers and the principals, between teachers and teachers, and between teachers and students. For instance, Grissom and Keiser (2011) found there is higher job satisfaction when there is teacher-principal race congruency. Elsewhere, others have found that teacher-principal gender congruence is also associated with teacher satisfaction and turnover (Grissom, Nicholson-Crotty, & Keiser, 2012).
Likewise, teacher-student race matching has also been found to influence teacher satisfaction, a strong predictor of teacher turnover (Engel, Jacob, & Curran, 2014; Fairchild et al., 2012;
Renzulli, Parrott, & Beattie, 2011; Stearns, Banerjee, Mickelson, & Moller, 2014). For instance, Fairchild et al. (2012) found that teacher-student race congruence is a positive and significant predictor of teacher satisfaction. In short, recent research in relational demography suggests that teacher-principal race/gender matching and teacher-student race matching can influence teacher attrition.
Accountability. In terms of external programs from the district or the state that rely or call upon accountability at the teacher or school level, there has been much development in state and federal programs and initiatives that aim to make changes to the teacher labor markets to attract qualified teachers and retain qualified and effective teachers. An additional part of the
logic model of accountability is that some attrition is probably good –in particular, attrition among low-performing teachers. Murnane and Steele (2007) noted some proposed policies, such as the use of teacher evaluation and teacher merit pay, to increase the supply of effective teachers and how they can be distributed more equitably. However, many of these policies were newly enacted, and they varied widely in terms of implementation and management. Moreover, their effectiveness was still being assessed and debated by 2006 (Murnane & Steele, 2007). Recent work has highlighted how schools and school personnel respond differently to teacher evaluation, how evaluation is framed differently depending on the contexts, and that teacher evaluation can have a direct impact on teacher satisfaction, commitment, and retention (Marsh et al., 2011; Murphy, Hallinger, & Heck, 2013; Peterson, 2000; Weiss, 1999).
Relatedly, there are policies and programs that link teacher evaluation, most often via student performance, with consequences and rewards. A couple of examples of these new programs are the Teacher Advancement Program and the Teacher Incentive Fund. The Teacher Advancement Program was established in 1999, but there was not an evaluation of its effect on the teacher labor market until after 2005 (e.g. Glazerman et al., 2013; Glazerman & Seifullah, 2012; Springer, Ballou, & Peng, 2014). The Teacher Incentive Fund, initially a $600 million federal grant established by Congress in 2006, was expanded and supported as part of the American Recovery and Reinvestment Act (ARRA) in 2009. At least two studies assessing the effectiveness of the TIF funds have been published recently (Slotnik et al., 2013; Wellington et al., 2016). Moreover, there are other state merit pay programs have been implemented and evaluated in the last 10 years (Pham, Nguyen, & Springer, 2017).
Additionally, there has also been evidence that federal programs and initiatives may influence the teacher labor market, especially teacher retention decisions (Brownell, Bishop, &
Sindelar, 2005; Harrell et al., 2004; Hill & Barth 2004). Lastly, recent works have found accountability of school administrators, most often the principals, can also influence teacher satisfaction, commitment, and attrition (Boyd et al., 2011; Grissom, 2011; Stockard & Lehman, 2004). Recently, using longitudinal administrative data, Grissom and Bartenan (2017) have found plausibly causal estimates of principal effectiveness on teacher turnover. In sum, teacher evaluation, teacher merit pay, federal policies such as NCLB, and principal effectiveness must be considered in how teacher retention and attrition is conceptualized.
School improvement. There is a long history of school reform efforts and over the last several years there has been a swell of new approaches and evaluations of school improvement efforts, including comprehensive school reforms and research-practice partnerships, that have a strong focus on teacher development and leadership (Borman, Hewes, Overman, & Brown, 2003; Bryk, Gomez, Grunow, & Lemahieu, 2015; Coburn & Penuel, 2016; Cohen-Vogel, Cannata, Rutledge, & Socol, 2016; Datnow & Castellano, 2001; Rowan, Correnti, Miller, &
Camburn, 2009). These school improvement developments aim to increase teacher buy-in and develop teacher capacity as teachers and leaders (Nguyen & Hunter, 2018; Rubin, Nguyen, &
Cannata, 2017), which theoretically and empirically can incentivize teachers to stay in their schools (Guarino, Santibañez, & Daley, 2006; Macdonald, 1999; Shaw, 2016). However, even though school reforms have been around for decades, there has not been any rigorous analysis of the effects of school improvement efforts on teacher attrition and retention until very recently (Heissel & Ladd, 2017; Sun, Penner, & Loeb, 2017). The recent implementations and evaluations of school reforms and research-practice partnerships may affect the teacher labor market and should contribute to the scholarly understanding of teacher attrition and retention.
Work force. Work force is a category of determinants that comes mostly from the employee turnover literature (Cotton & Tuttle, 1986; Griffeth, Hom, & Gaertner, 2000;
Rubenstein, Eberly, Lee, & Mitchell, 2017). These determinants are factors that reflect the employment opportunities inside and outside of teaching and policies that can influence attrition and retention at the district or state levels but not related to accountability or school improvement efforts. They include employment rate, accession rate, teacher salary, non-teacher salary, late hiring, retention bonus, union and tenure. The employee turnover literature indicates that the overall employment rate and existence of alternative job opportunities generally influences whether people stay or leave their current occupation (Griffeth, Hom, & Gaertner, 2000), and these determinants can logically be extended to teacher attrition and retention (Barbieri, Rossetti,
& Sestito, 2011; Clotfelter, Ladd, & Vigdor, 2011). Relatedly, the accession rate or the percent of new employees added during a hiring period, usually one year, could also affect attrition decisions (Cotton & Tuttle, 1986). Late hiring is also another factor that may relate to teacher attrition (Cotton & Tuttle, 1986; Jones, 2011). In terms of monetary incentives, non-teacher salary, teacher salary3 and teacher retention bonuses could incentivize or disincentivize teachers from leaving the profession (Griffeth, Hom, & Gaertner, 2000; Rubenstein, Eberly, Lee, &
Mitchell, 2017). Lastly, tenure reform is expected to affect teacher attrition and retention (Goldhaber, Hansen, & Walch, 2016).
In sum, there are clear theoretical reasons or empirical evidence that these recently developed categories, relational demography, accountability, school improvement, and work force, may influence teacher attrition and retention. All together, these nine subcategories grouped under personal, school, and external correlates form my conceptual framework of
3 Salary was moved to the work force category instead of school resources since salary is mostly set at the district
teacher attrition and retention. This current study extends and improves the seminal studies by Guarino, Santibañez, and Daley (2006) and Borman and Dowling (2008) by using improved methodological systematic search, and expanding the search from 1980-2005 to 1980-2017, adding in more than a decade of research and development, particularly in light of recent advances in new categories of determinants of attrition and retention as well as the proliferation of longitudinal data systems. The updates to the findings and to the conceptual framework, guided by theories and supported by empirical evidence, should enhance the scholarly understanding of what influences attrition and retention and provide the most up-to-date comprehensive review of the field’s knowledge on teacher attrition and retention. In the next section I discuss my systematic review process of the empirical literature on teacher attrition and retention and how the data are evaluated.
Data and Methods
This study is designed to examine the determinants of teacher retention and attrition, or the factors that have been found to explain why teachers persist or why they leave the teaching profession. To define the eligibility criteria, literature search, data analysis, and reporting conventions, I follow the Preferred Reporting Items for Systematic Reviews and Meta-Analysis standards as defined by Moher et al. (2009).
Eligibility Criteria. Following Guarino, Santibañez, and Daley (2006) and Borman and Dowling (2008) and expanding on their inclusion criteria, the primary studies eligible for inclusion in this systematic review need to meet the following criteria: (a) the sample is comprised of teachers in K-12 education; and includes (b) characteristics of individuals who leave or remain in the teaching profession; (c) characteristics of schools and districts related to
teacher attrition and retention; (d) compensation policies such as teacher merit pay programs that may affect teacher retention and attrition; (e) pre-service and in-service policies that affect teacher retention; or (f) other characteristics or factors that are related to teacher retention and attrition. As noted previously, this study endeavors to include studies that employ long-term longitudinal data that can capture the dynamic teacher career trajectories and studies that evaluate state and federal programs and initiatives that aim to change the teacher labor markets.
Additionally, this study focuses on the determinants that influence whether teachers stay or leave the teaching profession and not on whether they switch schools.
Literature Search. I obtained primary studies from searching commonly used economic and general social science databases, including ERIC, WorldCat, ProQuest, JSTOR, NBER and EconLit. Through an iterative process, I created the following search string: teacher AND (attrition OR turnover OR retention OR leav* OR suppl* OR career OR attitudes OR mobility OR commit* OR persist*). I also searched for “grey” literature using Dissertation and Thesis Repositories in WorldCat and ProQuest as well as a general Google search for evaluation reports of well-known merit pay programs such as the Teacher Advancement Program (TAP), the Teacher Incentive Fund (TIF), and the Texas District Awards for Teacher Excellence Program (DATE).4 In addition to searching databases, my literature search included an examination of reference lists and previous reviews of the teacher retention and attrition literature (Borman &
Dowling, 2008; Guarino, Santibañez, & Daley, 2006; Johnson, Berg, & Donaldson, 2005;
Wilson, Floden, & Ferrini-Mundi, 2001).
Studies Meeting Eligibility Criteria. Starting with the results returned from my search of databases and previous reviews, I used a two-phase process to screen for primary studies that meet all eligibility criteria as illustrated in Figure 1. First, I read the title, abstract, and
introduction for all studies obtained in my original search. I retained a study if the title, abstract or introduction mentioned that the study contained empirical results pertaining to teacher retention and attrition. Search results can be found in Table 2. In total, I screened 23,187 studies.
Table 2: Results by database
Database Results
ERIC 5,233
WorldCat 4,445
NBER 4,270
ProQuest 3,895
DOAJ 2,614
JSTOR 1,111
Google scholar 1,000
Taylor and Francis online 619
Total 23,187
Search string: Teacher AND (attrition OR turnover OR retention OR leav* OR suppl* OR career OR attitudes OR mobility OR commit* OR persist*)
In phase two, I was left with 286 studies for full text reading where I assessed whether each study fits the eligibility criteria outlined above. From these studies, I excluded studies due to lack of relevant teacher outcomes, non-empirical results, and duplicate reports. For multiple reports from the same study (e.g., a dissertation and corresponding journal article or reports from multiple years for the same evaluation), I kept only the most current publication. After screening, I was left with a sample of 152 studies that met all eligibility criteria.
Coding Reports. A second coder and I coded relevant information for each of the 132 eligible studies using an improved taxonomy based on Borman and Dowling (2008) to include new factors that have been found to relate to teacher retention and attrition such as teacher- principal race match. The coding schema and descriptions are provided in Appendix Table 1.
Any discrepancy was resolved by consensus between the two coders.
Results
I present the results of my systematic review of the literature here for each category for personal, school, and external correlates. In comparison to Guarino, Santibañez, and Daley (2006) and Borman and Dowling (2008), the numbers of studies for each category have increased substantially and in many cases, I am able to provide a more nuanced picture than was previously possible.
Personal Correlates
Teacher Characteristics. Across more than a dozen studies, when age is operationalized as a continuous measure, older teachers are generally less likely to exit the system than younger teachers (Donaldson & Johnson, 2010; Kukla-Acevedo, 2009; Nah, 2015; Rees, 1991; Sass et al., 2012). Relatedly, when age is operationalized as a binary comparison of those who are 30 years of age or younger compared to those who are older than 30 years, I observe that older teachers are less likely to leave (Clotfelter, Ladd, & Vigdor, 2011; Gritz & Theobald, 1996;
Imazeki, 2005; Ingersoll, 2001; Ingersoll & May, 2012). The findings are comparable when age is operationalized as a binary of 28 years of age or younger (e.g., Boe et al., 1998). Taken together, these results suggest that many young teachers leave the system within a few years of entry into the profession. In terms of gender, in contrast to the two previous systematic reviews, I find that the female teachers are not more likely to leave the system, particularly with the newer studies and studies with longitudinal data (Barbieri, Rossetti, & Sestito, 2011; Boyd et al., 2011;
Grissom, Nicholson-Crotty, & Keiser, 2012). Across these new studies, researchers find that female teachers are just as likely to leave as male teachers. It may be that how gender influences attrition has changed over the last ten years, or that studies with longitudinal and more recent
data are able to provide a more accurate representation of the influence of gender on attrition than prior works.
In terms of race, new data have allowed me to compare the likelihood of leaving the profession for minority teachers as a whole, and Black teachers and Hispanic teachers individually compared to White teachers. While previous reviews have found that all minority teachers are less likely to leave teaching (Borman & Dowling, 2008; Guarino, Santibañez, &
Daley, 2006), I find that minority teachers, broadly defined as non-White teachers, and Black teachers are not more likely to leave than White teachers, particularly with results from more recent studies (Clotfelter et al., 2008; Djonko-Moore, 2016; Goldhaber, Gross, & Player, 2011;
Harrell et al., 2004; Kelly & Northrop, 2015; Smith & Ingersoll, 2004). However, across many studies, Hispanic teachers are less likely to leave teaching than White teachers (Adams, 1996;
Kukla-Acevedo, 2009; Moore, 2011; Newton, Rivero, Fuller, & Dauter, 2011; Sass et al., 2012;
Dagli, 2012). These results indicate that retention rates are similar across most racial groups, although Hispanic teachers are more likely to stay in teaching than other groups.
In regards to marital status, married teachers are not more or less likely to leave teaching than non-married teachers (Harrell et al., 2004; Kukla-Acevedo, 2009; Rees, 1991;
Stinebrickner, 2002). Contrary to popular belief, teachers with young children are not more likely to leave teaching (Arnold, Choy, & Bobbitt 1993; Boe et al., 1998; Harrell et al., 2004;
Stinebrickner, 1998, 1999, 2002). The last few determinants in teacher characteristics are related to the school. We find full-time teachers are significantly less likely to leave teaching compared to part-time teachers (Arnold, 1993; Beaudin, 1993; Jones, Maier, & Grogan, 2011; Smith, 2006;
Smith & Ingersoll, 2004). Moreover, the more satisfied the teachers are with their teaching career, the less likely they are to leave the profession (Cannady, 2011; Kelly & Northrop, 2015;
Renzulli, Parrott, & Beattie, 2011; Dagli, 2012). Lastly, as the distance between where teachers live and where they teach increases, the more likely they are to leave the profession, although there are only two studies on which to base these conclusions (Barbieri, Rossetti, & Sestito, 2011; Steele, Pepper, Springer, & Lockwood, 2015).
Teacher Qualifications. Teacher qualifications is one of the most well studied areas of teacher attrition. First, teachers with more academic abilities as measured by GPA or test scores such as the SAT or the ACT are slightly more likely to leave than teachers with less academic abilities (e.g., Clotfelter, Ladd, & Vigdor, 2011; Goldhaber, Gross, & Player, 2011; Perda, 2013).
Similarly, teachers who come from more competitive or most competitive colleges, as defined by Barron’s, are more likely to leave teaching than those from less competitive colleges (Boyd et al., 2011; Clotfelter, Ladd, & Vigdor, 2011; Erickson, 2007), but there are some mixed findings (Rickman & Parker, 1990; Kelly & Northrop, 2015). In other words, most studies find that teachers with more academic abilities as measured by test scores or the college they attended are more likely to leave than those with less academic abilities. In terms of graduate degrees, there are many mixed and insignificant findings for teachers with graduate degrees compared to those with only undergraduate degrees or without any degrees (Djonko-Moore, 2016; Harrell et al., 2004; Imazeki, 2005; Kelly & Northrop, 2015; Newton et al., 2011; Ondrich, Pas, & Yinger, 2008). A study on National Board certification provides causal estimates that teachers with National Board certification in North Carolina are more likely to leave the state than teachers without the certification, which the author attributed to leaving the state but remaining in the profession elsewhere (Goldhaber & Hansen, 2009); conditional on remaining in the state, Goldhaber and Hansen find teachers with the certification tend to move from schools with high levels of minority students to schools with lower levels. Similarly, there are mixed findings about
the effect of being highly qualified as defined by NCLB and having internship experience (Connelly & Graham, 2009; Goldhaber, Krieg, & Theobald, 2016; Luke, 2014; Moore, 2011).
However, across several studies teachers who have regular or standard certifications are much less likely to leave teaching than those who do not (e.g., Harris-Mcintyre, 2013; Helms- Lorenz, van de Grift, & Maulana, 2016; Ingle, 2009; Johnson & Birkeland, 2003; Kelly, 2004;
Luke, 2014; Newton et al., 2011), although there is some evidence that teachers in some alternative training programs such as Professional Development Schools may be more likely to stay in teaching than traditionally trained teachers (Latham, Mertens, & Hamann, 2015; Latham
& Vogt, 2007).
The results also indicate that teachers who teach in certain subjects are harder to retain.
For instance, teachers in STEM and special education subjects are more likely to leave the profession than those who teach other academic subjects (Cowen et al., 2012; Grissmer & Kirby, 1992; Ingersoll & May, 2012; Kirby, Berends, & Naftel, 1999; Ogundimu, 2014; Stinebrickner, 1998, 1999). This speaks to the difficulty schools and districts have in retaining STEM and special education teachers (Billingsley, 2004, 2007; Ingersoll & May, 2012; Sass et al., 2012).
Lastly, studies consistently find that new teachers in the first three years of teaching are more likely to quit teaching than veteran teachers (Boe et al., 1998; Clotfelter et al., 2008; Djonko- Moore, 2016; Jackson, 2012; Steele et al., 2015), and that teachers with prior non-teaching career experience are not more or less likely to leave teaching (Boyd et al., 2011).
School Correlates
School Organizational Characteristics. There have been many studies looking at the relationship between school organizational characteristics and teacher attrition. These
characteristics range from enrollment and urbanicity to professional development and mentoring.
In terms of student enrollment, studies generally find school size is not an important factor (Mont
& Rees, 1996; Kelly, 2004; Imazeki, 2005; Goldhaber, Cross, & Player, 2011; Jones, 2011). In contrast to previous reviews (Borman & Dowling, 2008; Guarino, Santibañez, & Daley, 2006), recent studies do not consistently find that urban teachers are more likely to leave teaching than rural teachers (Bradley, Green, & Leeves, 2006; Donaldson & Johnson, 2010; Imazeki, 2005;
Jackson, 2012; Kelly, 2004; Moore, 2011; Smith, 2006). On the other hand, studies generally find that high school and middle school teachers are more likely to leave teaching than elementary school teachers, although many results are statistically insignificant (Marso & Pigge, 1997; Smith & Ingersoll, 2004; Stinebrickner, 1998). In terms of school sector, in agreement with previous reviews private and charter school teachers are more likely to leave teaching than traditional public school teachers (Hahs-Vaughn, 2008; Redding & Smith, 2016; Sass et al., 2012; Stinebrickner, 1998; Stuit & Smith, 2012). Perhaps unsurprisingly, teachers who teach at hard-to-staff schools or schools with low stay ratios may be more likely to leave than those teaching at schools with high stay ratio (Goldhaber, Krieg, & Theobald, 2016; Ronfeldt, 2012).
Though the first half of the school organizational characteristics determinants provides some understanding into teacher attrition, the second half of these determinants may also provide policy relevant insights. Studies generally find that teachers are more likely to leave schools with higher student disciplinary problems (Djonko-Moore, 2016; Ingersoll & May, 2012; Kelly, 2004;
Kraft, Marinell, & Yee, 2016). Relatedly, schools with better work environment as characterized by better facilities and less school problems also see less teacher attrition (Boyd et al., 2011;
Buckley, Schneider, & Shang, 2005; Ingel, 2009; Loeb, Darling-Hammond, & Luczak, 2005;
Martin, 2011; Moore, 2011; Stuit & Smith, 2012). Teachers who have stable teaching
assignments are also more likely to be retained (Ost & Schiman, 2015). Along this line, across several studies, teachers are less likely to leave teaching in schools with stronger administrative support (Boyd et al., 2011; Brown & Wynn, 2008; Eller, Doerfler, & Meier, 2000; Ingersoll, 2001; Luke, 2014; Smith, 2006; Urick, 2016) or with higher levels of teacher collaboration or cohesion (Boyd et al., 2011; Fuller, Waite, & Irribarra, 2016; Kraft, Marinell, & Yee, 2016;
Smith & Ingersoll, 2004). Comparable to administrative support, beginning teachers who experience induction and/or mentoring are also less likely to leave teaching than those without (Cannady, 2011; Luke, 2014; Mihaly, Master, & Yoon, 2015; Smith, 2006). This is particularly true for teachers whose mentors who have previous experience working in the school (Rockoff, 2008). Relatedly, teachers who indicated they had good in-service professional development are less likely to leave (DiGaudio, 2017; Erickson, 2008). Teachers with higher classroom autonomy are also less likely to leave than those without (Ingersoll & May, 2012; Dagli, 2012).
Surprisingly, there are mixed reports of the influence of leadership on teacher attrition. Jackson (2012), Kraft and colleagues (2016) and Ladd (2011) find that teachers’ reports of high levels of principal leadership or influence reduces the likelihood of leaving. On the other hand, Boyd et al.
(2011) and Dagli (2016) find that teachers’ reports of high levels of their own leadership and influence increases likelihood of leaving, and in contrast, Shaw (2016) finds that participation in a teacher leadership program can positively help keep teachers in the classroom.
Taken together, these results suggest that there are many school organizational characteristics that could be used to lower teacher attrition. In particular, lowering student disciplinary problems, improving work environment, increasing administrative support, providing better professional development and induction/mentoring for beginning teachers and
affording teachers more classroom autonomy are all viable actions that can be taken to reduce teacher attrition.
School Resources. Even though the relationship between school resources and student achievement has been well studied, the relationship between school resources and teacher attrition has not been well attended to. First, there are mixed findings on the relationship between the school expenditure for support per teacher and teacher attrition (Eller, Doerfler, & Meier, 2000; Gritz & Theobald, 1996; Imazeki, 2005; Kirby, Berends, & Naftel, 1999). Similarly, providing classroom assistants or teacher aides, or reducing classroom size does not seem to reduce the likelihood of teachers leaving the profession (Barbieri, Rossetti, & Sestito, 2011;
Eller, Doerfler, & Meier, 2000; Feng, 2010; Gritz & Theobald, 1996; Smith & Ingersoll, 2004).
However, providing adequate teaching materials to teachers does seem to matter (Gritz &
Theobald, 1996; Loeb, Darling-Hammond, & Luczak, 2005; Smith, 2006; Stevens, 2010).
Student Body Characteristics. The most significant development in this area over the last ten years is the study of how student achievement is related to teacher attrition. Across several studies, scholars have generally found that teachers are less likely to leave teaching in schools with better student performance (Boyd et al., 2008; Eller, Doerfler, & Meier, 2000;
Hanushek, Kain, & Rivkin, 2004; Loeb, Kalogrides, & Beteille, 2012; Newton et al., 2011; West
& Chingos, 2009). While the result may not be novel, it is robust to increases in student average test scores as well as comparing high to low performing schools.
What may be surprising is that, while teachers may leave teaching more in high minority schools than low minority schools (Hahs-Vaughn & Scherff, 2008; Dagli, 2012), increases in percent Black students, percent Hispanic students, or percent minority students are not consistently associated with increases in teacher attrition (Boyd et al., 2011; Feng, 2009;
Hanushek, Kain, & Rivkin, 2004; Goldhaber, Gross, & Player, 2011; Gritz & Theobald, 1996;
Imazeki, 2005; Ingle, 2009; Kelly & Northrop, 2015; Loeb, Darling-Hammond, & Luczak, 2005;
Newton et al., 2011; Smith, 2006). These findings suggest that, as a percent increase, the influence may be too small to detect or that the relationship between percent minority and teacher attrition is not a linear relationship. Relatedly, percent increase in free- and reduce-price lunch, FRPL, does not seem to be highly correlated with teacher attrition (Boyd et al., 2008;
Feng, 2009; Gritz & Theobald, 1996; Hansen, Backes, & Brady, 2016; Kelly & Northrop, 2015;
Moore, 2011). However, percent increase in FRPL may not adequately capture the difference between schools in poverty conditions and affluent schools. When poverty is measured differently, such as the majority of students with low socioeconomic status, some studies do find that teachers are slightly more likely to leave teaching in high poverty schools than low poverty schools, but almost all of the findings are insignificant (Eller, Doerfler, & Meier, 2000; Fulbeck, 2014; Luke, 2014; Stinebrickner, 1998, 1999; Dagli, 2012). Lastly, percent increase in IEP/LEP is not highly associated with increases or decreases in teacher attrition (Djonko-moore, 2016;
Falch & Ronning, 2007; Feng, 2010; Ingle, 2009; Moore, 2011). In short, there is only weak evidence that student body characteristics significantly influence attrition with the exception of the level of student achievement.
Relational Demography. This category is a recent and ongoing development of how teacher-principal, teacher-teacher and teacher-student race or gender congruency influences teacher attrition. In terms of teacher-principal race congruence, Grissom and Keiser (2011) and Harris (2007) find that teachers are less likely to leave teaching when they are of the same race as the principal. To a smaller extent, teachers are also less likely to leave teaching when they are of the same gender as the principal (Grissom, Nicholson-Crotty, & Keiser, 2012). Similarly,
